TWI270263B - Fast joint detection - Google Patents

Abstract

K data signals are transmitted over a shared spectrum in a code division multiple access communication system. A combined signal is received and sampled over the shared spectrum. The combined signal has the K transmitted data signals. A combined channel response matrix is produced using the codes and impulse responses of the K transmitted data signals. A block column of a combined channel correlation matrix is determined using the combined channel response matrix. Each block entry of the block column is a K by K matrix. At each frequency point k, a K by K matrix Lambda(k) is determined by taking the Fourier transform of the block entries of the block column. An inverse of Lambda(k) is multiplied to a result of the Fourier transform. Alternately, forward and backward substitution can be used to solve the system. An inverse Fourier transform is used to recover the data from the K data signals.

Description

1270263 发明, DESCRIPTION OF THE INVENTION: TECHNICAL FIELD OF THE INVENTION The present invention generally relates to a wireless communication system. In particular, the present invention relates to data detection in a wireless communication system. [Prior Art] The first picture is a description of the wireless communication system ίο. The communication system 1 has a base station 12, which is connected to user equipment (ue) 14 to 14 (14). Each base station 12 has an associated operating area and is here in communication with a User Equipment (UE) 14 in its operating area. In some systems, such as: frequency division duplex (Z/Plus/CDMA) using code division multiple access and time division duplex using code division multiple access (TDD/CDM magic, multiple communication systems in the same spectrum) Transmission is performed. These communication systems are distinguished by their channelization code. For more efficient use of this spectrum, time division duplex (10)/drain communication systems using code division multiple access are used for segmentation into communication. The repeat of the slot. One of the communications in this system will be assigned a single or multiple associated code time slot. Since multiple communications may be transmitted on the same spectrum and in the same time, in this system the receiver must be able to identify multiple communication rooms. The means of detecting such signals is multi-user detection (10). In the multiple user details (10)), the signal associated with all user equipment (UE) 14, that is, the user, is detected simultaneously. Detected by - single-transmitter - another means of multi-code transmission 1270263 is a single-use test (SUD). In a single-user sd (sud), in order to restore data from the multi-code transmission of the receiver, the received signal is de-spreaded by a singularization phase and using a single or multiple code (one (10). The means of user detection (MUD) and the single-user-side (SUD) equalization phase include: using -Cholesky or - approximating chQlesky decomposition. These methods have a southern complexity. This high degree of complexity will This leads to an increase in power consumption and causes a shortened battery life of the user equipment (UE) 14. Therefore, the object of the present invention is to provide other means for detecting the received data. [Summary of the Invention] The K data signal is in a code. One of the multiple access (CDMA) communication systems transmits on a shared spectrum. A combined signal is received and sampled on the shared spectrum. The combined signal has such K transmitted data signals. A combined channel response matrix is used. The data code and impulse response of the K transmission data signal are generated. A block field of a combined channel correlation matrix is determined by using the combined channel response matrix. Each block item is a KxK matrix. At each frequency point k, a κ χ κ matrix Λ(1) is determined by the Fourier transform of the block item that calculates the block field. This ΚχΚ matrix Λ(1) is inversely matrix multiplied by one of the Fourier transforms. Alternatively, forward and backward substitution systems can be used to solve the system. An inverse Fourier transform system is used to restore data from such data signals. 1270263 [Embodiment] The second figure introduces the use of a code Multiple access time division duplex (τιφ/ CDMA) g enables speed joint detection - to simplify the wheeler 26 and the receiver 28, although the fast joint side can also be finer than its (four) system, such as · using code division multiple Frequency division duplex (CDMA/FDD) for access. In the "system", a transmitter 26 is located in each user equipment (UE) 14, and the multiple transmission circuit 26 for transmitting multiple communications is located at each base. The unit 12 can be located in a base station 12, in the user equipment 14, or both. The transmitter 26 transmits data on a wireless radio channel 3. In 26 A data generator 32 generates data to be communicated with the receiver 28. The wheat/revolution/spike sequence inserting device 34 utilizes an appropriate data code to spread the data and distribute the reference material to the appropriate time slot. One of the intermediate text training sequences is time multiplexed to generate one or more communication bursts. A typical communication burst 16 series has an intermediate text 2 〇, a guard period 18, and two data fields 22, 24 As shown in the third figure, the intermediate text 2 is used to separate the two data fields 22, 24, and the guard cycle 18 separates the communication bursts, so that the transmissions from different transmitters 26 are considered. The arrival time difference. The two data fields 22 and 24 contain the information of this communication burst. This (equal) communication burst is modulated into a radio frequency (sand) using a modulator 36. An antenna 38 is passed through the radio channel 3 to radiate the radio frequency (i.e.) signal to one of the antennas 4 of the receiver 28. The type of modulation used for transmission communication can be any known to those skilled in the art. _ 'such as: four-phase shift keying ((4)) or quadrature amplitude modulation (QAM) 〇 this receiving antenna 40 series Receive various RF signals. These received signals are demodulated using the -Exploration 42 to generate - the base frequency. The baseband signal is sampled by a -sampling means 43' such as a single- or multiple analog-bit digital converter, or the like, at a one-time wafer rate or a multiple-multiple wafer rate. These samples are processed in the slot at this time via, for example, a channel estimation device 44 and a fast joint detection device 46, using appropriate data codes assigned to the received bursts. The channel estimation device 44 uses the intermediate text training sequence elements in such baseband samples to provide channel information such as channel impulse response. All channels that transmit signals can respond as a matrix. This channel information is used by the fast joint detection device 46 to estimate the transmission data of the received communication burst as a soft symbol. The fast joint detection device 46 uses the channel information provided by the channel estimation device 44 and the known diffusion data code used by the transmitter 26 to estimate the data expected to receive the communication burst. Although the Fast Joint Detection is interpreted as a basic communication system using the 3rd Generation Partnership (3Gpp) Universal Terrestrial Radio Access (UTRA) Time Division Duplex (TDD) system, this fast and bonded test can also be applied to other systems. The system is a direct sequence wide frequency division multiple access (W-CDMA) system in which the uplink and downlink communication systems are limited to mutually exclusive time slots. This receiver 28 receives all of the bursts that arrive at the same time. These solid bundles 1270263 overlap above each other in an observation interval. For the 3rd Generation Partnership Project (3GPP) Universal Ground Radiation Access (10) A) Time Division Duplex (Lion) system, the individual bay fields of the time slot are gamuted to the observation interval. The data code of the kth New Zealand is not awkward. The κ bursts can originate from κ different transmitters, or less than 不同 different transmitters for multiple data code transmission codes. Each of the communication bursts (4) stock field stores a predetermined number, Ν, the pass symbol. Each symbol is transmitted using a predetermined number of wafers, which is a diffusion factor, SF: thus each data field has _ wafers. After each of the wireless radio channels, the individual symbols have an impulse response, such as an impulse response of a length cargo wafer. Length w of the Wei Weipu, 57. To this end, each of the field fields has a length of one of the wafers or NC wafers. The κ field system of each of the κ data fields in an observation interval can be modeled at the receiver using equation (1): Κ Equation (1) £(k) = Α(1)d(k), k^ l, is the receiving contribution of the kth field + the combined channel response of the kth field. Prepare N Hao array. In the a sigh of the fourth (1) of the fourth element symbol ringing - zero padding version. This symbol is a convolution of the k-field estimation response h(1) and the diffusion of the field. ") is an unknown data symbol in the data field of the second data. The h system has a length w wafer and can be represented by the equation (2). : (k) = X(k) Equation (2) .hw , No. X(1) is the reaction transmission H gain and road surface consumption. £, pure channel impulse response! 27〇263 For uplink communication, each ^ and each work (1) are different. For the downlink, all fields The domain system has the same but different (1) different systems. The right transmission diversity is used in the downlink, and all the transmissions of all the fields and the different channels (3) are transmitted on the wireless channel. The system is based on the equal r = ir + n - ~; Equation (3) S is a zero-average noise vector. 00 Combine all the data fields into an overall response _ level to combine the unknown data of each cluster To the whole capital, «, equation (1) becomes equation (4) -. Ι = Α - + Π equation (4) use - human-machine system engineering (MMSE) solution to determine the basis according to equation (5) (four) '(f) equation (5 (.)喺 represents the Hermetian function (complex conjugate transpose). _ Human System Engineering (MMSE) R is based on equation (6). R=M+rI Equation (6) π-system noise Typically, this channel is used to estimate the device material and the unit matrix is used. Fast Fourier Transform (FFT) is used, although other Fourier transforms can be used, but this equation is best solved according to equation (7). [F(d) 〕k Equation (?) 1270263 The needle is slightly longer than the 立立观_. [.]_ indicates that this equation is solved at each frequency. Λ + square object turns λ the big block of the project. The derivation of the cube diagonal matrix 说明 is described as follows. Discarding directly to solve equation 7, equation (7) shoots the forward and backward # generation method. The fourth graph uses the fast joint detection to determine the data direction. The method-process engraving of the combined channel response moment _ is determined by using the estimated noise and the spread data code (4) of each burst c (1), 48. Through this combined channel correlation matrix, divination,

49. At each frequency point, a matrix is determined by calculating the Fourier transform of a block of squares, 5Q. Preferably, the present embodiment uses a center line that is at least W lines from the left or right side of the matrix R.

F[A〇 k is a fast Fourier transform (fft) of one of the matrix multiplications, and the inverse matrix of each matrix Λ(1), [A']-, is determined. In order to determine [F (this)] k 〔 Λ ^] -1 and F [AHr] k are multiplied at each frequency point. Alternatively, [ρ (this)] k is determined using LU decomposition. λ(1) is decomposed into a lower triangular matrix [and an upper two-dimensional matrix U, 52. Use forward substitution, [F (Αν)], 53, and backward substitution, U[F(d)]k, 54, [F(d)] potential. The g system is determined by inverse fast Fourier transform using one of F (this), 55. The derivation of equation (7) is explained below. One of the minimum mean square error (MSE) solutions of equation (4) is determined according to equation (8). Although equation (7) is based on the solution of a human machine system engineering (MMSE), the rapid joint detection system can be implemented using other means, such as a zero force means. 12 1270263 R- (A.A+a^I)d=Av #式(8)—If the -zero force solution is used, the H term is omitted from equation (8), such as: (AA) Α γ. The following description is one of the derivations of this human-machine system (e.g.), although a similar derivation can also hide-zero force solution. For the purpose of introduction... Simplify the example R (where, and W = 2) according to equation (8). This example can be extended to any point of view. 0 0

And *ί and I 〇^ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 〇^1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ^ 0 0 0 0 0 0 0 0 0 0 0 Equation (9)

The size of this matrix R is usually (KNS) X (KNS). In each of the items in this matrix R, R is a KxK square. Within the dashed line of the matrix R, the matrix is cyclically looped, that is, one of the cyclic matrices extends in the direction of the square. The part of the matrix R, the part of the non-square loop, depends on the maximum multipath extension spread, W. In equation (9), one of the matrices of this matrix R is cyclically extended, R. , according to the equation (10) ° 13 1270263 E: 5. 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 βΓ 0 0 0 0 0 -3⁄4 Equation (10) One-digit Fourier transform (DFT) - similar to f, the matrix D is determined, such as r = DNDh. One such matrix D is based on the equation (11 D: •1 know jl&E - ...e M, βύχ _ ΛΤΓ τ layer·· 4? r 1 uj ilCfac he ......£ )4ύπ Jjis^c \ ......£ Σκ Equation (11) L is a KxK unit matrix. The product DhD is based on the equation (9) Equation (12) is a unit matrix of mxm. This block is a matrix of the matrix R. This is the multiplication of the moment, such as: Equation (13). 14 1270263 RD: (13) + horse + + ^+^) + + is)

(V i^. J** jl&c iit jlSc Ί5Γ j'ifa + Α<Ρ Wa + now<? Wa ) +S^eI + jlSfc + ^£~ +ή£~) jlflc jilk

iiiL il^. ji Know Wa + Atf Wa (deduct jlSt+i^ iifc. wa +v 乐j+ S^£ ^ ) (for £ && + heart% j?^c jlSDc + R^e Wa +^ Λ ) jlD8c jl^lfc (V Wa +i2,^ Wa +^f jhi slave jlgfk+ ^ +J^£ tfa ) jl your equation RJ) each item is a KxK box. A square diagonal matrix Λ system root (4) ((4) "rushing" ΛΑ equation (14) matrix Λ size system (KNs) x (KNs). The matrix Λ each Λ (η according to equation (15) 〇 # Λ Equation (15:) Λ(1) The instrument has a square block and has a 1 Η non-zero item. This matrix D is multiplied by this matrix Λ, such as: according to the equation (π)

DA Λ(1) J4* j4* Λ I _ j8* jl&c · Λ,丨VTiT jl&c ~ £nr Λ[丨) 心15Γ l^*bc ·· Λ Ί jitic himinr Λ(1) Λ(1) Λ嘹)\ Ύχ 13⁄43⁄4 J Λ Ι [~Ι · ·. Λ阳Ί .,, 丨 j jl-ark i^H^£ Wa jM4c Equation (16) 15 1270263 Various items of matrix D shown in equation (16) One block. The system of equations generated by the columns equal to the matrix RJ) and the columns of the matrix D is fixed. Therefore, the same set of equations is generated by equalizing any column of the matrix and the same column of matrix D. To introduce equation (13), the first column of the matrix RD is based on the equation (π). [(R〇+RH+R2H + R, + R2) ^ (R〇e--+RHe--+Rae--^ + R,e-) ^ (R〇e--+Rfe-+R^ , (R〇0j18 "Ns+RHe36·+R2He54"Ns+R2e· a+Rej 180/r/Ns )] Equation (17) The first column of matrix D is based on equation (18) [Λ (1), Λ (1) ej2"Ns, Λ (3) ej4"Ns,. · · 5 y^cNs-ngj, Λ )^^jl8;r/Ns Equation (18) Equal to the two columns of items, you can get the equation ( 19) and equation (2(8) Λ(1)=(R+RJRM + R + R) Equation (19) Λ(1)e (Re_+RHe (10) + RHe— + Re -external + re ν ) =e— (R〇+RHe—+ RHe—+ R2e-—+ Re_j2^/Ns Equation (20) Therefore, L is based on the equation (21). Λ(ζ)= (R〇+R1Hej2^5+ RHej^Ns+ R2e-^+ R,e- Equation (21) Similarly, Λ-" is based on equation (22). J2(N卜2) "Ns 八(四)一(R.+ RHei2 (4)"ns + RHe40^2) "Ns + (4)" Ns + rg Equation (22) 16 1270263 Λ(10) is based on equation (23). 八(10)=(R#+R,He recorded (10) + RHe,,,R2e_/external R,e Equation (23) Although the equation (17) to Equation (23) introduces the use of the matrix r£d and the first column of da' but any column can be used to determine Λ(1). Use a central column, the (Ns/2)th column (or the fifth column of equation (7)), and Λ(1) according to the equation (19) 〇Λ(η= (Ro + RH + RH + R + RO etc.) Equation (19) • Equations (19) through (23) are fast Fourier transforms (pyp) of κ' K squares. Since these squares are multiplied by a scalar index, this step is called a, Block Fast Fourier Read (FFT) ''. Typical means of calculating Fast Fourier Transform (FFT), such as:

Matlab software function fft, which is a fast Fourier transform (FFT) for calculating a single-edge sequence. Since each eight (1) system is a bilateral sequence, the calculation of λ can use a Fourier transform function fft {〇, 〇, ..., R, R, R, RH, rh, ..., 〇, 〇} and multiply it by a suitable exponential function of a central column, such as: according to the equation read (27), implemented. , "中中 p = [ceil (M/2) - 1] / conditional equation (27) ^7) to equation (27), calculate all eight (1) systems can use the matrix R - early one Therefore, the matrix does not need to be determined. The matrix Ε Φ can be directly derived Λ (). Preferably, the real relationship uses at least one of the four sides of the distance matrix R, because these columns have - Group complete R. 17 !27〇263 Using Λ(1) and matrix D' square circulant matrices iu can be re-expressed as equation (28) and equation (29) 〇 equation (28) equation (29)

RcD=DA

Rc= (l/K) (DAD〇 matrix D and Λ are respectively size (ΚΝ) χ (KR) Since D»D=NSLS'D—(i/n ) Dh, the equation can be obtained (3〇 Rc-i-Ns ( (D〇(D) = Ns ( (D/Ns) Λμ (Dh/n )]

Equation (30) This Human Machine System Engineering (MMSE) solution is based on equation (31). Sa R—! (A〇 equation (31) The data vector detected is the size (orphan) xl. This human-machine system engineering (MMSE) solution is based on equation (32). D«d=Ai (D« (A〇) Equation (32)

The matrix system size (KNs) x (KNs) has 1 [> 1 square, and the inverse matrix of the matrix 根据 is based on equation (33). W r , 1 w r Γ Λ Λ -1-. Λ (4)· — - μ (4) 广 Equation (33) This inverse transformation requires an inverse matrix of one of the matrices Λ (1). Therefore, this data vector 4 is determined according to equation (34). 18 !27〇263 f F (4)]=[U ''[F (Ay)]k Equation (34) β (34) can be closed at times - times and multiples of wafer f 'such as: twice The chip rate is sampled (4) of the received signal. For a receiver with a multiple wafer rate, the moment corresponding to the complex bit rate is in the same form as equation (9), which approximates a block cycle.

In order to reduce the complexity of F (A's complexity, this embodiment may use a fast Fourier transform (4)) which is advantageous for structure a. Structure a has a structure that approximates a block cycle. However, the structure records a non-square matrix with a size of (10)), (RK). An introduction to one of the matrices A is based on equation (35). Song (〇) "to ΨΚΙ" [m ant (1) W\2) to) & 2l a V) (o) 俨 (〇, 狮呦(〇) 〇0 ή[η(〇) bn〇) 35)

Each bi (k) (i) is the channel response h(1) and the spread data code c (the stack of k corresponds to the first user of the jth wafer interval at the ith symbol interval. B (·), where each block is represented by the brackets in equation (35), and equation (35) becomes equation (36). 0 Hall 0 0 臞鼸· 0 0 Face 臞 0 5 (2 5(1) B(O) :: :" Equation (36) An egg) 0 〇^(Ό B(O) 〇5(2) B (1) 5(0) = ; : 〇0 ...

19 !270263 As shown in the above formula, the extension of the matrix A is marked as 4 by the extension of the block cycle. The flat core cycle A matrix can be cut into three matrix scores according to equation (37). A=Di Λ .D2H 荨 (37), is a (10) (four) Zheng car, Λ, is a square diagonal matrix of a size (NsSF) x (Li). This cube diagonal matrix Λ has a solution ((4) the same form. Missing, matrix Λ various items Λ, (1) according to equation (38) series - SM box f, Λ, sleep in in ^ in in 〜f is in the equation (38) n \〇Ό J 秘 and the matrix D in equation (11) has the same form ^ is the form of the equation (evil).

Ύι j±l J e ISF e ^ ISF I jISc

D,: J is called) 4ύκ jlSx _ g Wa T _ g Wa J m j i gy _ This is only i!

jlBfik g""ST

L system - SFxSF unit matrix Equation (39) The product of the singularity of IX and the suffix 'form' B(1) and 卜' is formed according to equation (10)). 20 !27〇263 I chat ‘啊丨醉^印卢丨卢··丨嗍^ ΒΠ)# ‘

Antelope N 吟 N· 如 J > » *

犹 · I • · 1 哗 pm 作 阼

AcD2 — Equation (40) The size of the 尬 (NsSF) x (RK), and the size of each block is sf, k.

In the multiplication matrix and the eight temples, the form, (i), the product system is formed. &Λ size system (MSF) X (M), and the size of each block is sFxK. Comparing the matrix AJM and the matrix & Xiang Xiang in the same column, you can get the equation (41). Nine (1) = [B (0) + B (1) + B (2)], eight ' (2) = [ Β ( 0) + Β (1) e, /Ns + Β ( 2 ) e-i4&quot ;Ns], A!(Ns-·1) =[B(〇) +B (1) e_i,2)"Ns +B (2) e-j4(Ns-2)"Ns], A. (Ns) = (B(0) + B (1) e-j2(Ns-,); r/Ns+B (2) 0-i4(Ns-l) ff/Ns ^ Equation (41) Therefore, Each Λ, (k) can be determined using one-sided sequence of one of the (SFxK) blocks. Using equations (38) & D2HD2 = NSL, how do you get equations (42), (43), and equations? (44) Equation (42) Equation (43) Equation (44)

A=Di Λ iDJJ ΑϋΗ (D ugly) DJi (A〇=Ns[ΛΉ (DH )] 21 1270263 Therefore, [F(A〇]k is determined according to equation (45) using Fast Fourier Transform (FFT) [F (A〇]k=Ns[ Λι〇〇]"[F (I)]k equation (45) Similarly, since matrix A is approximately square-loop, r=A„a+ σ2ΐ can also be used. The fast Fourier transform (FFT) of Λ is calculated. To reduce the complexity, the inverse matrix of each Λ(1), [Λ(1)]-1, can be performed using lu decomposition. Each [Λ〇 一 (ΚχΚ) matrix, its LU decomposition According to the equation (46) 〇A(i)=Ll^ ^ ^ ^ ^ ^ Equation (46) L is a triangular matrix, and u is an upper triangular matrix. Equation (7) is based on the equation (47) And equation (48), using forward and backward substitution methods. [(k)][f (Ay)]k Equation (47) [Ag (k)]H[F (3)]k, etc. (48) For the ruling, in order to improve the position difference (BER) of the data symbols at the end points of each data field 22, 24, the sampling from the intermediate text part 2〇 and the carettage period 18 is used for the first The information shown in Figure 5 is in the middle. (d) All samples of the last symbol in the yard field are used to determine [the sampling system extends the wafer to the intermediate text 20 and the caret period 18 (the length of the impulse response). This extension considers the generalization of the last symbol of this field. In the data field 丨22, these sampling systems are used to spread the dice to the Chinese character. This intermediate text is in the data detection section. In the data field 2 24 and 1270263 words, these sampling systems are extended to the caretaker cycle 18 - the wafer. ...Special 快 稍 立 叶 FFT (FFT) silk needs - the length of the field is silk Analysis. These fast-following Fourier transform (FFT) implementations—the main factor-difference method (PFA). This main factor algorithm (pFA) implementation requires this field length to be - the main number, such as: sixty - For the convenience of the main algorithm (10)) Fast Fouriera _ implementation, the conversion of the good extension of a predetermined main liver algorithm (10)) storage. Such as the fifth riding, f material field] and data field 2 system extension P chip to the main factor algorithm you want (p(4) length Or, the sixty-symbol block fast catching Fourier transform (10) is extended to a length of sixty-four square fast Fourier transform (FFT), which requires a private fast Fourier transform (four)) calculation. The performance of the array is reduced, and its performance is usually improved. ((K D K)/2) [2 (SF + W - 1) - υ - ((Κ 2ΙΟ / 2) (SF + WD, where , ϋ + W-1) / SF) +1) Equation (49) The calculation of the fast joint gamma is based on the following. # a之叶 The complexity is KxSFxW. The computational complexity of the calculation M is based on the equation (^.

〕 (ru/2) (R, ((SF-matrix vector multiplication system calculation (a〇m is obtained, which has - complex 陋(sf-υ. Calculate the fast Fourier transform (fft) of the ith row of the moment) [X (R1.) calculation is required. Calculate a Fourier transform system f tear (N, _) calculations. The inverse matrix of each matrix [h], without decomposing 嶋olesky 23 1270263 * K3 calculations are required. For R frequency points, the total number of calculations is thief 3. Playing [F(4)]k=[Λ(1)]H[F(A,)]k requires [multiplication (for Ns frequency points) Therefore, the total number of calculations is κ. The inverse fast Fourier transform (FFT) of [F (έ)] requires K (N1〇gN) calculations. To introduce the complexity of fast joint detection, processing time-sharing Duplex (TDD) 甘 曰 λ Τ ί ί ί ί ί ~ ί ί ί ί ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Matrix Α, (5), matrix, and vice versa: ghost, [Λ;]-, followed by ^

Calculation of Av, F[Αν], [F(4): turn: calculation_cluster execution... (four): every == turn: = touch matter know _.峨梅绍在第一量_ 丛丛-calculation A MR0PS calculation Μ 3.0 Calculation F (j 4.4 Lu ten nose [Λ(1)],, 9. 2614 12. 4928 Each _ 丛发勃 t MROPS Calculation Ay Calculation F [Ατ] 28.11 2· 3154 1270263 Calculate F[〇n]dA small [F(AT)]k VIII 3.1232 F[(d)] inverse fast Fourier transform 2.3154 Fast joint detection required all 65. 0182 ... Bai Wan The number of real numbers (MR0PS) is the first note: in the first table, (8) is directly calculated as a -matric vector multiplication. The right LU knife solution is used to determine [A(k)]_, then complexity The system is reduced to % 8678 million real operations per million (10) PS). If Fast Fourier Transform (fft) is used to determine (8), the complexity is reduced from 65._ million real operations per second (send s) to 63. 9928 million real operations per second (failed pS). The complexity comparison of fast joint detection and other detection techniques is described below. For the time division duplex (TDD) burst type 丨 (in other words, the complexity of the following three techniques is based on the second table.

&Ά MRQPS Approximate Cholesky Basic Joint Detection (JDChol) 82. 7 Single User Detection: Following a Hadamard Transformation Basis Diffusion (SDChol) Approximate Cholesky Basic Equalization 205. 2276 Fast Joint Detection (JDFFT) 65.0182 25 1270263 The second type of _Technology and - Reference Matching Filter (MF) f The performance of the material side technology is based on the simulation of more than (10)_. Butterfly _ use the precision provided by the software athb, that is, does not consider the limited precision effect. These simulations use wideband code division multiple access (w-_) time division duplex (10)) fourth group

(10)) The scales are 6 wins 8 and 12, and are executed without the heterogeneity of the downlink, so as to facilitate comparison with the single-user side. As shown in the sixth and seventh figures, for the first and third cases, the performance of the fast ear (four) speculation (IDFTT) is often close to 0^16^ basic joint sampling (JDChol) . The performance of other data side methods is not as good as __based joint predicate (JDChol) or fast joint detection (constrained τ). For the second channel of the Time Division Duplex (TDD) izg group (WG4) shown in Figure 8, the Fast Joint Debt Test (JDFFT) appears to present a certain comparison with the Cholesky Basic Joint Detection (jDChol).

Some recessions. This also occurs in the Single User Detection (SUD) based Ch〇lesky algorithm (SDChol). For a high data rate service, such as: a 2 Μ _ service (as shown in Figure IX to Figure 11), the performance of fast joint detection (jDpyp) is close to or slightly less than Cholesky basic joint detection (jDChol) ), but better than other methods. [Simple description of the diagram] 26 1270263 The first diagram is a wireless communication system. The second figure is a simplified transmitter and a fast joint detection receiver. The third picture is an introduction to a communication burst. The fourth diagram is a flow chart of one of the preferred embodiments of fast joint detection. The fifth picture shows one of the data bursts in the extended processing area. Figures 6 through 11 show graphical representations of the simulated performance of fast joint detection of other data detection methods. [Component_Touch] _ 10: Wireless communication system 12, I2j_~12§ · Base station 14, 14~14: User equipment (dish) 16: Typical communication burst 18: Guardian Zhou Jia: Intermediate text 22, 24: Data field · 26: Simplified transmitter 28: Receiver 30: Wireless radiation channel 32: Data generator 34: Modulation/diffusion, sequence insertion device 27 1270263 36: Modulator 40: Antenna 42: Demodulation 43: sampling device 44: channel estimating device 46: fast joint detecting device

Claims (1)

1270263 picking up, claiming patent scope: 1·- type code multiple access user equipment, the flipping device is connected to a κ data signal above the shared frequency, the user equipment comprises: receiving and sampling - having Means for combining the combined signals of the material transfer over the frequency spectrum; means for generating a combined channel response using the data code and impulse response of the 4K data; ... using the combined channel response matrix to determine a block of one of the interactive relation matrix blocks, Each block item of the block line is a κ*κ matrix; (10) a device for blending money sampling without a matrix of one of the complex conjugate transposes of Fourier transform; A device for generating a Fourier transform of the data vector; and means for calculating an inverse Fourier transform of the data vector Fourier transform to generate data of the data signal. 2. A multi-access base station of the code division, the base station receives a data signal above the common spectrum, the base station includes: receiving and sampling - a combined signal having a data above the shared spectrum The device uses the data code and the impulse response of the data to generate a combined channel response device; 曰 'Use the touch button _ _ decision - the device of the interactive light matrix block 29 K70263, the block Each block item is a κ viewing matrix; a device that multiplies one of the combined channel response matrices of the combined signal samples to perform a Fourier transform; a Fourier transform of each of the square items - means for inversely converting the double-leaf transformation of the bribe to produce the (10) vector-Fourier transform; and calculating the means for recording the data of the transfer signal. 3. If you apply for a full-time syllabus, the financial calculation of the conversion of the vertical leaf is to use the combination of the commencing channel (four) _ the _ record _ combined signal, and calculate the total complement multiplication - the result of the - Zhuang miscellaneous. 4. If the application is the second torn base station, one of the block projects of Caijing Corner Scale is used to determine the data. 5. If the base station mentioned in the second paragraph of the patent application is applied, the data decision occurs in the data-domain __ of the time-division duplex communication burst, and the combined signal sampling system extends beyond the data field. Time period. 6. The base station of claim 5, wherein the extended sampling of the combined signal samples extends beyond the data field time period to a length of the impulse response. • For the base station described in Shen. 5, the combined signal sampling system is extended over-the-counter: _ domain _ period, __ combined signal-length becomes non-compatible with a master sub-method fast Fourier -length. 30 1270263. The user equipment of claim 2, wherein the Fourier transform system calculates the common signal transposition by multiplying the common transposition of the combined channel response matrix by the combined signal and calculates the conjugate One of the results of transposition multiplication is a Fourier transform. 9/ If an application is made to refer to the timing device, one of the blocks of the pair of cubes is used to determine the data. Ίο. As described in the application of the scope of the patent range i, wherein the data is determined to occur in the time-sharing duplex communication burst--Lin field time, the female combined signal sampling system extends beyond the data field Domain time period. Port π. As claimed in the patent ship _ the device of the immortal, wherein the combined signal sampling extends and extends the length of the overdue data field. 12. The user equipment of claim 10, wherein the combined sampling system extends beyond the data field time, and the ticks are t-gate periods, and the length is one of the combined signals. Become the -_ sub-algorithm fast catching Fu Li _ compatible - length. 1270263 柒, designated representative map: (1) The representative representative of the case is: (2). (b) The representative symbol of the representative figure is a simple description: 26: simplified transmitter 28: receiver 30: wireless radiation channel 32: data generator 34: modulation/diffusion/training sequence insertion device 36: modulator 40: Antenna 42: Demodulator 43: Sampling device 44: Channel estimation device 46: Fast joint detection device 捌 In the case of a chemical formula, please disclose the chemical formula that best shows the characteristics of the invention: